How much should a trader rely on backtesting results?

A strategy that performs well on historical data may struggle when market conditions change or when execution costs are included. Backtesting is useful, but it has limitations. What additional evidence would you want to see before trusting a strategy with a live or funded trading account?